We will build upon the research accomplished during two very successful previous funding cycles. The particular aspects of HIV biology, which we propose to study, have been motivated by pressing needs of researchers designing HIV-1 vaccine candidates. These needs include a better characterization of viral dynamics during transmission and early infection, a deeper understanding of the interplay between host immune response and viral evolution, and a continued development of computational methods that make full use of ultra-deep sequencing (UDS) data. This project will combine our proven expertise at modeling and computational analysis of rich sets of sequence, clinical, and assay data available from completed and ongoing research projects. First, we will design, implement, validate, and disseminate computational methods for identifying and predicting sequence and structural HIV-1 envelope epitopes targeted by monoclonal neutralizing antibodies and polyclonal sera. Computational analysis of neutralization takes a fraction of the time and cost of traditional molecular approaches and has the potential to match or exceed the accuracy of the latter. Predicted epitope residues or motifs can be subsequently used for rapid targeted in vitro validation, cataloged, and made publically available, thereby greatly accelerating the pace of rational HIV-1 vaccine design. Using these predicted and validated motifs, available sequences can be screened to estimate the proportion of circulating viruses neutralized by a given nAb. Second, we will develop biologically realistic evolutionary models for the comparative analysis of HIV-1 sequences that account for natural selection, strong biases in amino-acid substitution patterns, recombination, heterotachy and host effects. The proposed comparative sequence analysis methods will determine molecular mechanisms of viral escape from host immune response, adaptation during and following transmission, compartment-specific dynamics, and the co- evolution of viral gp160 and host IgG heavy and light chain sequences. Understanding these dynamics is crucial to the design of immunogens that elicit protective immune responses. Third, we will create HIV-1- specific bioinformatics tools and methodology that leverage UDS data for advancing our understanding of viral evolution and population dynamics. UDS technology allows for an unprecedented amount of sequencing granularity in complex samples, such as HIV-1 populations isolated from infected individuals. A reliable estimate of the mutational spectra and distribution of common and rare genetic variants of HIV-1, combined with statistical models, will yield better estimates of within-host population dynamics, and shed light on the interplay between genetic drift and selective forces experienced by the virus. A key deliverable of the project will be open-source software tools that address the unique statistical analysis, interpretation and data access needs of this and similar projects working with HIV and other rapidly evolving pathogens.